OHBM Oral History Interview Series: Bruce Rosen

Professor Bruce Rosen

By Nils Muhlert

Bruce Rosen is a physicist and radiologist who, for the past 30 years, has been instrumental in the introduction and development of functional MRI. Bruce currently serves as the director of the Athinoula A. Martinos Center for Biomedical Imaging at the Massachusetts General Hospital. Here, we found out about the exciting early stages of putting a team together to discover and develop the principles of fMRI, and helping to found OHBM in the process. 

---

Nils Muhlert (NM): I'm here today with professor Bruce Rosen, for the OHBM oral history initiative and to celebrate 25 years of OHBM.

Bruce Rosen (BR): So a big event.

NM: Yeah. I'd like to start by asking about your background and how you became interested in neuroimaging

BR:  Somewhat by serendipity. I was working with colleagues at what was then the MGH NMR centre now the Martinos centre working on various technical issues and a colleague of mine, a chemist, actually brought a problem to me. He was developing contrast agents to look at the liver. The game there was to try to develop agents that would then get picked up by the liver and would create contrast enhancement, the way we would today with our gadolinium agents. This was, of course, at the very dawn of NMR back in the early 80s, so there wasn't that much known in designing an agent that would be specifically taken up by the liver. But they found that when they acquired the very first images, rather than showing up bright, which is what they were hoping for like a standard contrast enhancement, the images actually showed up dark.  The images then subsequently brightened up. And so he said, you know, what's up with this? It set me to thinking about the problem. And I came up with a hypothesis, that when the agent was first being introduced into the blood, rather than a relaxation effect, the agent caused the images to enhance. This was the so-called susceptibility effect, where there are differences in the magnetic field locally around blood vessels, that led to the signal drop. Then, as the concentration in the blood vessels increased, you would get the realisation of that. This was easier to observe in the brain, rather than the liver where this washout happened very quickly. 
 
By good luck, I had a young neurologist who was doing a research fellowship with us, his name is Arno Villringer, also one of the fMRI pioneers, he was looking for a project. He was doing lactate imaging in the brain with me, but that project was going slowly. So we hit upon the notion of studying the brain because the blood brain barrier would keep the contrast agents inside the blood vessels longer and allow us to explore the effects. So with Arno, we started out a series of experiments to show that it was this magnetic susceptibility mechanism. He then had quite a well cited paper at the time - the first technique to measure a functional attribute of the brain, in this case, the blood flow delivery, which you couldn't see with conventional T1 effects because the agents didn't leak out. But in a normal brain, you saw this enhancement of this T2 effect, this susceptibility effect. That introduced me to the notion of the brain and perfusion. After that paper was published, Arno went back to Germany, but we had a new grad student come into the group: Jack Beliveau. 

Jack was especially keen on studying brain function. He introduced the notion that if you could image with this perfusion technique multiple times, you could compare differences in brain state, much the way PET scanning had done previously with their oxygen measurements. We then had the good fortune to have early access to very high speed imaging, so called echo planar imaging, which at that time was not available on the commercial scanners. There was a small company, serendipitously started by my former thesis advisor who had worked for Sir Peter Mansfield, that was located in Boston called advanced NMR. They were looking for a site to prototype this instrument, and decided to site it in our hospital. Jack was saying, let's do these experiments. Peter Fox then lent us the visual goggles that he had used for his first PET experiments (we used literally the same goggles; he had them shipped up from Texas). With those goggles we were able to demonstrate that we could show visual activation with MR. That was the first fMRI experiment, Jack's famous cover of Science picture and the rest is history.


NM: So that leads on to another question, which is “what contributions are you most proud of in your career?”

BR: That's not a bad one (laughs). I think my greatest contribution was just cheering on the fabulous people that I had the opportunity to work with. Arno, Jack, Dave Kennedy, Robert Weisskoff at the time. Subsequent to that Peter Bandettini was in our lab. Randy Buckner was a fellow in our lab. Anders Dale was a fellow in our lab. Many of the people that today are the leaders in the field. Chantal Stern was another. Many of them came to the lab when we first made these discoveries. Kenneth Kwong who did the first BOLD experiments, started as a fellow. Maybe the thing I'm proudest of is creating the environment where these really smart people could come and explore and pursue their own passions. Jack had a passion for studying the brain. My job was to help create an environment where these people could do their best work and then turn them loose, and of course, Jack in particular did great things.

NM: And, and what do you see happening with neuroimaging in the US nowadays?

BR: It's not ever really lost its lustre, but I think neuroimaging in the US has gotten yet another wind. This is largely because of the US BRAIN Initiative.  This is somewhat similar to the efforts that are happening elsewhere around the world, in Europe and China, etc. The US effort has been focussed on the cells and circuits level, but there is increasingly an interest in human neuroscience. And that's, of course, where functional imaging really has come to the fore.

We're able to study humans in ways that you can't do in rodents. To look at individual cells and circuits, you have to come up with non-invasive ways to study those principles in humans. The Brain Initiative in the US has been generous in supporting human neuroscience, and neuroimaging is a very important part of that. That's led to an infusion of significant resources and technology, both in developing new instruments and developing new analytic methods. That feeds into more applications, where people can use those tools to study all sorts of new things. 

NM: Do you see more convergence now between the findings from circuits in animal models, and what we're seeing from human neuroimaging? 

BR: It's something I've been thinking about. I would say that that convergence is beginning. There's certainly people that have begun to take the tools that were developed for modern molecular neuroscience and apply them to questions of the fundamental mechanisms of neurovascular coupling. Work with folks like Anna Devor at UCSD, David Boas now at BU and others, have really taken advantage of those and made great efforts. 

There's also a convergence in the scales that we can study as neuroimaging continues to improve, in part because of these investments in the technology. I think we'll be able to move from the systems level down to the mesoscale. Meanwhile at the same time in animal models, the folks doing optical imaging, electrophysiology, are going up there, from arrays of hundreds to arrays of thousands and 10s of thousands, from small optical field of view to whole cortex. And they're also now observing and quantifying mesoscale phenomenon. 

I have this notion that, just like the Transcontinental Railroad, that started at both sides and met in the middle, as the animal folks build up from single cells to small local networks to mesoscale phenomena, and as the human neuroimagers work their way down, that we're going to meet at this mesoscale. And then I'm hoping that we’ll be able to learn and manipulate and study in animal models with optogenetics and all the rest of the modern tools will really inform what we can then see at the mesoscale directly in humans. This will then allow us to better understand what we're seeing at the systems level. That's what I'm looking forward to over the next five or ten years.

NM: You played a role in the creation of OHBM. What was it like in the early days?

BR: It was a lot of fun. There were a lot of interesting characters, most of whom were new to me. Most of those founding fathers (it was mainly fathers in that era rather than mothers but I’m glad to see that’s all changing) came from the PET world. Those were the people that did functional imaging back in the late 80s and early 90s. Folks like Jack and myself, we were kind of newcomers with this new thing: MR. In fact, I remember seeing the slides recently from Peter Fox's first meeting in Texas - I wasn't there, but by the second meeting, I'd somehow slipped in the door. And of course, that was shortly after Jack had published his paper and Ken had published his paper. So now suddenly, the old PET guard was paying a little attention - though somewhat skeptically and somewhat at a distance because they weren't tools they were familiar with. 

It was certainly fun in the beginning to get to know these people. And they had a lot of interesting personalities and a lot of their own perspectives on what should or shouldn't be done. But it was clear to us that MR was going to be the future. We felt that we belonged in the room, even if nobody else quite knew that at the time. 

After the organisation of the very first meeting, when the question came up about who was going to take on the mantle and hold the second meeting - that's when Jack actually stood up and said, we'll do it. Everyone said, "Great," and then of course, he turned to Dave Kennedy and myself and said, “Okay, now we have to do it”. So he kind of agreed to it before we ever had much say in it but in retrospect, I was very, very glad he did and he had a very able partner in Dave Kennedy, who really did a lot of the heavy organisational lifting. Jack was an inspirational figure, Dave was the guy that can actually turn the inspiration into action. I was there to underwrite it, in case we got into financial trouble I'd somehow bail us out. So we each had a role to play. 

That meeting took us from 800 odd people to 1200 people. That took the little budget that the folks from Paris were able to share with us and we handed off to the next meeting something like $150,000. So that put the whole basis on a firmer financial footing. And although in the history books, the third meeting is considered the first official OHBM meeting, actually the incorporation of the OHBM under that name actually happened to support the Boston meeting. We needed to create a corporation for that. So it actually was incorporated just prior to the Boston meeting. And those articles of incorporation were used for several years before we then created another purpose built group. 

It was a fun time and it was exciting for us still as kind of the new kids on the block to play a role. Having people come to us, that was probably the meeting where that shift of focus away from PET to MR really became clear. I think in the first meeting 25% of the meeting was about MR. By the Boston meeting it was probably already up to half and of course in subsequent meetings that grew tremendously. So, we again, got on the ground floor.

NM: Are there particular memories that spend your time attending the annual OHBM meetings?

BR: I remember the Paris meeting extremely well. I was like a kid in a candy shop, just learning so much. It was largely unfamiliar to me. Almost everything I was seeing was new, and the people I was meeting were also largely new, that was a lot of fun. The Rolling Stones were in town for that meeting; I went to that show - and that was a lot of fun too. That first meeting was great. 

Boston was a whole different thing because we were in the spotlight. It's one of those things that in the beginning and the days building up to it, and on the first day, you're just very, very tense and very, very focused. But as the meeting went on, it was clear that somehow we had managed to pull this off. We found the pins for the posters at the last second so everybody got their posters up, and people were enjoying themselves. We began to relax. I remember we had a ska band by the name of Big Scallop band; it was our kind of party and, to my mind, still the best musical event in the history of OHBM. A lot of people danced very late - in fact we paid the band to play an extra set and people were just having a great time. 

NM: And how do you see OHBM has changed over the years? 

BR: I'll have to admit, for various reasons, it's been a few years since I've attended this meeting. I was somewhat out of touch, anxious to see for myself what it was like. It's really migrated into a very interesting and unique niche as I see it. 

There was a time when OHBM overlapped with groups like the ISMRM, the big MR meeting because there was so much focus on fMRI - it was a new thing which led to new technology, so both organisations were pushing that. But over time, the ISMRM has become more focused on the hardware and technology purely for acquisition. 

It's clear from this meeting that people here at OHBM are really thinking about how to analyse the data: network analysis, very sophisticated mathematics and analysing and interpreting what we're seeing in terms of the underlying brain function. That's something that's outside the scope of other meetings, and in much more detail than you'll see at neuroscience, although we may get little smatterings of that. So that, to me, has been very exciting. I think some of the best sessions at this meeting are people that are doing very creative things with how they're thinking about the data. And of course, that then filters back into how you collect the data and lots of novel ways that we’ll continue to push in both societies and will help move the field forward as a whole.

NM: And the last question, where do you see the future of neuroimaging then, over the next 5-10 years. 

BR: I absolutely don't have a crystal ball on that. It's clear the technology hasn't finished growing. Even MR, which is mature technology, continues to improve. People are now talking about the first of our 11.7 Tesla magnets coming online. People are already planning on 14 Tesla human magnets. So big magnets, that's one direction. The other direction is tools: optical imaging has been used for quite some time, but those tools are really getting much better quickly. Things like MEG is about to undergo a dramatic improvement with the advances in atomic magnetometers. PET probably has at least an order of magnitude improvement in sensitivity awaiting it over the next three to five years. That's based on novelty of crystal designs and geometries, and the application of deep learning to the reconstructions. That will take us to an area where we'll have a little bit of swing back to the old PET days. 

In many ways the society moved almost completely away from PET. But I think especially in this setting of neuromodulation, which was a very hot topic at this meeting, from the analytic side, with the modern generation of PET cameras, we can anticipate people being able to directly measure these neuromodulatory effects dynamically in vivo. That's going to be extremely cool and again, will help connect the dots, between the different modalities. We've been banging on this multimodal drum, and the OHBM has been foundational in helping that. I see many of the other tools that complement MR coming up to play an even more active and engaged role, less niche supporting players and more central to understanding what the brain is doing.

NM: And there have been a few very interesting talks here in Rome on simultaneous PET/MR...

BR: We're extremely keen on that. For example, people have made studies of receptor binding with PET for a long time. But it turns out if you simultaneously measure function with MR and the receptor dynamics. You can measure things like receptor trafficking, receptor internalisation, baseline receptor occupancy, these kind of quantitative measures that are actually very hard to make even in animal models. Now we can make those measurements in humans and with increased sensitivity, I think really be able to make those studies dynamically. That's certainly something that we're very interested in and again, it's one of those domains where I think the PET and MR communities really have a lot to learn from each other . So it's an exciting future,

NM: It's been fascinating. Thanks very much for your time.

BR: My pleasure.